A novel method for removing thin clouds from single satellite image is presented based on a cloud physical model. Given the unevenness of clouds, the cloud background is first estimated in the frequency domain and an adjustment function is used to suppress the areas with greater gray values and enhance the dark objects. An image, mainly influenced by transmission, is obtained by subtracting the cloud background from the original cloudy image. The final image with proper color and contrast is obtained by decreasing the effect of transmission using the proposed max-min radiation correction approach and an adaptive brightness factor. The results indicate that the proposed method can more effectively remove thin clouds, improve contrast, restore color information, and retain detailed information compared with the commonly used image enhancement and haze removal methods.